Abstract: A method is for designing an accelerator for digital signal processing including defining a software programmable fully pre-laid out macro by pre-laying out with a fixed topology a control logic of the DSP accelerator to obtain a fully pre-laid out control logic. The method further includes defining a hardware programmable partially pre-laid out macro by customizing a configurable layout area, thereby mapping a computational logic based on computation kernels related to an application of the DSP accelerator. A partially pre-laid out computational logic is therefore obtained.

Abstract: A memory base cell stores a bit of information implemented from a regular and compact structure made up of multiple identical and replicated base elements, on the “sea of gates” model, in which the base element of the structure is a cell able to be configured with a minimum width in relation to the particular technology used. Such a cell includes a bistable element with an input node operatively connected to a writing data line of the memory base cell, and an output node operatively connected to a reading data line of the memory base cell. The bistable element also has a first inverter and a second inverter arranged in a feedback configuration with respect to one another between the input node and the output node of the bistable element.

Abstract: A memory base cell stores a bit of information implemented from a regular and compact structure made up of multiple identical and replicated base elements, on the “sea of gates” Model, in which the base element of the structure is a cell able to be configured with a minimum width in relation to the particular technology used. Such a cell includes a bistable element with an input node operatively connected to a writing data line of the memory base cell, and an output node operatively connected to a reading data line of the memory base cell. The bistable element also has a first inverter and a second inverter arranged in a feedback configuration with respect to one another between the input node and the output node of the bistable element.

Abstract: A method is for designing an accelerator for digital signal processing including defining a software programmable fully pre-laid out macro by pre-laying out with a fixed topology a control logic of the DSP accelerator to obtain a fully pre-laid out control logic. The method further includes defining a hardware programmable partially pre-laid out macro by customizing a configurable layout area, thereby mapping a computational logic based on computation kernels related to an application of the DSP accelerator. A partially pre-laid out computational logic is therefore obtained.